Research On The Bio-photon Emission Signals Of Wheat Kernels With Hidden Insects

Bio-photon emission is the common life phenomenon in biological systems. As the product of metabolism in life process, it is sensitive to the internal change of biological system and the external influence of the environment. Due to its advantages of rapid, sensitive, reliable and non-destructive, bio-photon analysis technology has been widely applied. Combined with current research status, the ultra weak photon emission of wheat kernels attacked by maize weevil is analyzed by different methods in this paper. The main work is as follows:Firstly, the data of ultra weak bio-photon emission of normal wheat and wheat kernels attacked by maize weevil was analyzed, and then analyze the temperature of water whether has significant influence or not. Fitting the original signal of ultra weak bio-photon emission of wheat kernels with double exponential curve, compared the curve of normal wheat with the curve of wheat kernels attacked by maize weevil, the results showed that the bio-photon emission intensity of the infected wheat was higher than that of the normal wheat under different water temperatures. BP neural network algorithm was used to classify the infected wheat and normal wheat, and the recognition rate reached 80%.In order to test the size of the random degree, the permutation entropy algorithm is proposed in this paper. Using the permutation entropy algorithm to research the ultra weak photon emission signals of wheat, the difference of the ultra weak luminescence signals of different levels of the infected wheat was detected. There is no significant difference in the random degree of the signal before and after water injection, the data length, the embedding dimension and the delay time can affect the effectiveness of the permutation entropy detection. The recognition rate is 75% when classified with the characteristic parameter of permutation entropy.Established model for the ultra weak photon emission signal of wheat time series model. A suitable model is established according to the characteristics of the signal sequence, and the validity of the model is tested. Finally, a better fitting model is obtained.The research of this paper involves different methods of data analysis. In order to lay a foundation for the establishment of non destructive testing model of the wheat, the time domain characteristic parameter, the randomness of the signal sequence and the time series model of the signal are described.